Re: [ClusterLabs] ocf:heartbeat:IPsrcaddr generated failed probe "[findif] failed" on inactive nodes
On 07/02/24 14:57 +0100, Adam Cecile wrote: On 2/7/24 09:49, Oyvind Albrigtsen wrote: On 07/02/24 09:35 +0100, Adam Cecile wrote: Hello, Crm_mon show these errors on my cluster, while everything is working as expected: Failed Resource Actions: * Default-Public-IPv4-Is-Default-Src probe on gw-3.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 49ms * Default-Public-IPv4-Is-Default-Src probe on gw-1.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 48ms * Default-Public-IPv4-Is-Default-Src probe on gw-2.domain returned 'error' ([findif] failed) at Wed Feb 7 08:02:31 2024 after 64ms I think pacemaker is unable to check default source address on node which are not currently owning the IP addresses, which is expected. However Default-Public-IPv4-Is-Default-Src is +INF colocated with public IP addresses, so I do not understand why such errors are generated on inactive nodes. This is the probe-action, which will check whether the resource has the expected status (e.g. stopped for nodes where it's not running). You can either setup another IP on the same network on the interface to avoid these errors, or setting cidr_netmask and interface might help. IPsrcaddr doesnt advertise the interface parameter, so you probably have to do e.g. "pcs resource update -f Default-Public-IPv4-Is-Default-Src nic=" to set it anyways, so findif will be able to use it. Thanks ! You got it, it was indeed related to that. I tried setting up "nic" but it told me the parameter did not exist so I guessed it was not possible. Is that normal to use "private" attribute with --force ? Nice. It should be announced, so --force is just a way to override it. I'll make a patch to add it, so the parameters are listed for others who might use it without another IP in the IP-range on the NIC. Oyvind Albrigtsen Here are some config extracts: primitive Default-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.1 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive IPSEC-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.2 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started primitive Public-IPv4-Gateway Route \ params destination="0.0.0.0/0" device=eth1 gateway=1.1.1.254 \ op monitor interval=30 \ op reload interval=0s timeout=20s \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive Default-Public-IPv4-Is-Default-Src IPsrcaddr \ params cidr_netmask=24 ipaddress=1.1.1.1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started colocation colocation-Default-Public-IPv4-Is-Default-Src-Default-Public-IPv4-INFINITY +inf: Default-Public-IPv4-Is-Default-Src Default-Public-IPv4 colocation colocation-Default-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: Default-Public-IPv4 Public-IPv4-Gateway colocation colocation-IPSEC-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: IPSEC-Public-IPv4 Public-IPv4-Gateway order order-Default-Public-IPv4-Default-Public-IPv4-Is-Default-Src-mandatory Default-Public-IPv4:start Default-Public-IPv4-Is-Default-Src:start order order-Default-Public-IPv4-IPSEC-Public-IPv4-mandatory Default-Public-IPv4:start IPSEC-Public-IPv4:start order order-Default-Public-IPv4-Public-IPv4-Gateway-mandatory Default-Public-IPv4:start Public-IPv4-Gateway:start Any hint would be greatly appreciated ! Best regards, Adam. ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] ocf:heartbeat:IPsrcaddr generated failed probe "[findif] failed" on inactive nodes
On 2/7/24 09:49, Oyvind Albrigtsen wrote: On 07/02/24 09:35 +0100, Adam Cecile wrote: Hello, Crm_mon show these errors on my cluster, while everything is working as expected: Failed Resource Actions: * Default-Public-IPv4-Is-Default-Src probe on gw-3.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 49ms * Default-Public-IPv4-Is-Default-Src probe on gw-1.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 48ms * Default-Public-IPv4-Is-Default-Src probe on gw-2.domain returned 'error' ([findif] failed) at Wed Feb 7 08:02:31 2024 after 64ms I think pacemaker is unable to check default source address on node which are not currently owning the IP addresses, which is expected. However Default-Public-IPv4-Is-Default-Src is +INF colocated with public IP addresses, so I do not understand why such errors are generated on inactive nodes. This is the probe-action, which will check whether the resource has the expected status (e.g. stopped for nodes where it's not running). You can either setup another IP on the same network on the interface to avoid these errors, or setting cidr_netmask and interface might help. IPsrcaddr doesnt advertise the interface parameter, so you probably have to do e.g. "pcs resource update -f Default-Public-IPv4-Is-Default-Src nic=" to set it anyways, so findif will be able to use it. Thanks ! You got it, it was indeed related to that. I tried setting up "nic" but it told me the parameter did not exist so I guessed it was not possible. Is that normal to use "private" attribute with --force ? Oyvind Albrigtsen Here are some config extracts: primitive Default-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.1 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive IPSEC-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.2 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started primitive Public-IPv4-Gateway Route \ params destination="0.0.0.0/0" device=eth1 gateway=1.1.1.254 \ op monitor interval=30 \ op reload interval=0s timeout=20s \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive Default-Public-IPv4-Is-Default-Src IPsrcaddr \ params cidr_netmask=24 ipaddress=1.1.1.1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started colocation colocation-Default-Public-IPv4-Is-Default-Src-Default-Public-IPv4-INFINITY +inf: Default-Public-IPv4-Is-Default-Src Default-Public-IPv4 colocation colocation-Default-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: Default-Public-IPv4 Public-IPv4-Gateway colocation colocation-IPSEC-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: IPSEC-Public-IPv4 Public-IPv4-Gateway order order-Default-Public-IPv4-Default-Public-IPv4-Is-Default-Src-mandatory Default-Public-IPv4:start Default-Public-IPv4-Is-Default-Src:start order order-Default-Public-IPv4-IPSEC-Public-IPv4-mandatory Default-Public-IPv4:start IPSEC-Public-IPv4:start order order-Default-Public-IPv4-Public-IPv4-Gateway-mandatory Default-Public-IPv4:start Public-IPv4-Gateway:start Any hint would be greatly appreciated ! Best regards, Adam. ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] ocf:heartbeat:IPsrcaddr generated failed probe "[findif] failed" on inactive nodes
On 07/02/24 09:35 +0100, Adam Cecile wrote: Hello, Crm_mon show these errors on my cluster, while everything is working as expected: Failed Resource Actions: * Default-Public-IPv4-Is-Default-Src probe on gw-3.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 49ms * Default-Public-IPv4-Is-Default-Src probe on gw-1.domain returned 'error' ([findif] failed) at Wed Feb 7 08:00:22 2024 after 48ms * Default-Public-IPv4-Is-Default-Src probe on gw-2.domain returned 'error' ([findif] failed) at Wed Feb 7 08:02:31 2024 after 64ms I think pacemaker is unable to check default source address on node which are not currently owning the IP addresses, which is expected. However Default-Public-IPv4-Is-Default-Src is +INF colocated with public IP addresses, so I do not understand why such errors are generated on inactive nodes. This is the probe-action, which will check whether the resource has the expected status (e.g. stopped for nodes where it's not running). You can either setup another IP on the same network on the interface to avoid these errors, or setting cidr_netmask and interface might help. IPsrcaddr doesnt advertise the interface parameter, so you probably have to do e.g. "pcs resource update -f Default-Public-IPv4-Is-Default-Src nic=" to set it anyways, so findif will be able to use it. Oyvind Albrigtsen Here are some config extracts: primitive Default-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.1 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive IPSEC-Public-IPv4 IPaddr2 \ params cidr_netmask=24 ip=1.1.1.2 nic=eth1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started primitive Public-IPv4-Gateway Route \ params destination="0.0.0.0/0" device=eth1 gateway=1.1.1.254 \ op monitor interval=30 \ op reload interval=0s timeout=20s \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s primitive Default-Public-IPv4-Is-Default-Src IPsrcaddr \ params cidr_netmask=24 ipaddress=1.1.1.1 \ op monitor interval=30 \ op start interval=0s timeout=20s \ op stop interval=0s timeout=20s \ meta target-role=Started colocation colocation-Default-Public-IPv4-Is-Default-Src-Default-Public-IPv4-INFINITY +inf: Default-Public-IPv4-Is-Default-Src Default-Public-IPv4 colocation colocation-Default-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: Default-Public-IPv4 Public-IPv4-Gateway colocation colocation-IPSEC-Public-IPv4-Public-IPv4-Gateway-INFINITY +inf: IPSEC-Public-IPv4 Public-IPv4-Gateway order order-Default-Public-IPv4-Default-Public-IPv4-Is-Default-Src-mandatory Default-Public-IPv4:start Default-Public-IPv4-Is-Default-Src:start order order-Default-Public-IPv4-IPSEC-Public-IPv4-mandatory Default-Public-IPv4:start IPSEC-Public-IPv4:start order order-Default-Public-IPv4-Public-IPv4-Gateway-mandatory Default-Public-IPv4:start Public-IPv4-Gateway:start Any hint would be greatly appreciated ! Best regards, Adam. ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/